Passlabs Amplifier Projects |
||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Construction Building the Aleph4 I had to be a little budget minded. You can do a lot of good shopping in surplus stores, especially for power supply parts. Using second-hand here will not affect the performance of the amp one iota and can save considerable $$ for those worried about these things. For example (in $AUS .... for USD$ simply divide by 2):
So for a small fraction of the "new" cost I had most of the power supply! OK, so the filter caps may not last 20 years, but hell ..... at that cost I regard them as disposable. I used a RF filter IEC socket for the mains. I mounted the power switch (dual 10A breaker) on the rear to minimize AC lead length and used an inline 5A fuse. I used Pi filters for the supply rails. Those readers who have not read Nelson's article on Power supplies should do so before building such an amp. I mounted the rectifying bridge on a heatsink for 2 reasons. First it would obviously generate some heat of its own, but also I had no idea of the dissipation rating of the 2 main heatsinks, so I didn't know if I would have enough. I had this lying around, actually it was the cut-off end of the sink I had used for the tops of the Son of Zen project. Turned-out it was the right width for my tower case design. Given I had a total of 320,000 uF to charge, the inclusion of power thermisors 1&3 was essential (unless you want to have to click the main breaker 2-3 times to get the amp turned-on). On power-up I get a healthy buzz from the toroids due to the high current drain, which decays exponentially over 5-8 secs as the caps charge leaving near silence. There is a discernable 50Hz hum from the amp, but you have to place you ear up against the speaker grill for this to be just audible. From the circuit, Nelson did not include fuses on the supply rails, only the mains. I think doing both is sensible and it cost me nothing except the time to cut the holes in the rear panel. I used 5A breakers as I had them available. I did not include the 75°C thermal protection switches. A bit lazy on my behalf. Given the subsequent alteration to the design (ie. 3 vs 6 FETs) I wish I had done this and may add one/two later. All parts are obvious, except for C1 across Q3. The listed value is '103'. Many have asked me "what is this?" I have assumed it is standard labeling for a film cap, ie:
Another point of confusion is the value of R29. This is different on the 2 circuit diagrams in the service manual. On questioning Nelson has stated, "the value of R29 has varied over the production life of the amp, personally I prefer the higher value", so I went for 620ohm. I have received a dozen e-mails asking if I matched these (procedure) - the answer is yes. I ordered 25 each of both the IRFP244 and IRF9610 from Newark (discount price and remember, I saved a lot on the power supply!). Those of you out there should think about "group purchase" to attain both discount and ability to match well. How close did I get these? The 9610s easily to the 0.01V and the 244s to 0.05V:
For interest the DC offset of the final amp is < 10mV. If you plan on using the small FET mounting PCBs you will need to solder the FETs in place prior to mounting them on their heatsink. This can be tricky because they need to be soldered in a perfect plane to allow final mounting (if they are not you may get compromised heat transfer). First bend the FET leads at 90° at the boundary where the lead changes width. If you have a narrow pair of "long-nose" pliers you can grab all 3 leads at once in a straight line. This allows for a neat result and effectively shorts the 3 leads preventing accidental static to the gate. I used 2 pieces of acrylic and 2 small clamps. Place the FETs in a line along the edge, place the second piece on top so the FETs are "sandwiched". Align the FETs so they fit through the PCB holes and are straight, then clamp in place. Turn the sandwich over then solder in place. I noted on the Aleph4 stuffing diagram that the 9610s were given heatsinks, which were not present on the Aleph3 diagram even though the input bias current was the same. Alas, as I had started designing the PCBs on the A40 format and the 9610s were too close together to mount proper heatsinks, however you can get small "snap-on" heatsinks which are better than nothing and these are what I used. Clever folk may want to redesign the PCBs to allow for mounting heatsinks and I may do this one day if I have time. On running the amp the input FETs are hot, but certainly within limits and can be touched albeit briefly. The remainder of the construction is straight-forward for anyone likely to attack such a project. I followed Nelson's input wiring arrangement right down to the shorting plug in the XLR socket. I used gold-plated RCA sockets and speaker terminals. Wiring was with appropriate gauge OFC wire. The central earth point is a 2mm thick, 150x80mm piece of aluminium, isolated from the mains with a thermistor as described. The base, top, front and back I had made in 1.6mm mild steel by my local metal shop, this cost around $50AUS and is a lot easier than trying to bend sheet-metal yourself, unless you have access to the correct tools. I spray painted this myself, but given it is now living inside I may have this stripped and have the front/top powder-coated for a more professional finish - as I did with my A40 chassis. Have fun ....... |